Journal Logo

Special Communications: Methods

Prediction of ˙VO2peak from submaximal cycle ergometry using 50 versus 80 rpm


Author Information
Medicine & Science in Sports & Exercise: February 1997 - Volume 29 - Issue 2 - p 268-272
  • Free


Submaximal exercise on a bicycle ergometer has been used to predict mode specific maximal oxygen consumption (˙VO2peak) since the work of Astrand and Rhyming (3). Several researchers have extended this work(10,11,13,15,18); however, no study has examined the role of cycling cadence in the accuracy of the estimation. In fact, almost all studies have used the same cadence, i.e., 50 rpm.

Early researchers found that a moderate cycling cadence (33-60 rpm) results in a lower oxygen consumption than does cycling at lower or higher cadences(4,8). This economy of effort may be the reason that the developers of submaximal fitness tests chose to use a cadence of 50 rpm. However, most recent studies have shown that a higher cadence is most economical, particularly among individuals with cycling experience when exercising at higher intensities (6,12,19). A cadence of approximately 80 rpm is optimum based on these studies. This conclusion raises doubts about the nearly universal selection of 50 rpm for fitness testing purposes. Further, a common observation during fitness testing is that some subjects find 50 rpm to be uncomfortably slow.

We sought to determine the utility of using 80 rpm as the cadence in fitness testing. Our two hypotheses were that: 1) for a wide range of subjects, submaximal cycle ergometry tests performed at 80 rpm would be equally valid in predicting ˙VO2peak as those performed at 50 rpm; and 2) for subjects with cycling experience, 80 rpm would have greater validity than 50 rpm.


Subjects. All subjects were between 18 and 40 yr of age and apparently healthy as defined by the American College of Sports Medicine(1). For the purposes of this study, subjects were considered to be experienced (EXP) if they had engaged in cycling exercise(indoor or outdoor) for at least 30 min, 3 times per week over the past 3 months. For subjects to be considered nonexperienced (NEXP), they must have performed no cycling of any kind in the 3 months immediately prior to the study and must have reported never cycling on a regular basis at any time in the past. All subjects provided informed consent in accordance with institutional guidelines for research with human subjects.

A total of 65 subjects initially participated in the study. Those who did not meet the criteria for maximal exercise on both incremental tests were excluded from data analysis, leaving 30 EXP subjects (16 males, 14 females) and 28 NEXP subjects (15 males, 13 females). A summary of their characteristics is presented in Table 1.

Protocol. Each subject's height, weight, and skinfold measurements (the latter to estimate percent fat (16)) were recorded. Subjects were fitted with a mouthpiece (Hans Rudolph, Kansas City, MO), and ECG electrodes were placed in a lead II configuration. Each subject performed two incremental exercise tests on an electrically braked and calibrated bicycle ergometer (Sensor-Medics model 800, Yorba Linda, CA), one test at 50 rpm and one at 80 rpm. The seat was adjusted to provide a slight bend (approximately 5°) in the knee at full extension. Pedal straps were used. The tests were performed in random order and were separated by approximately 1 wk. Subjects were asked to abstain from alcohol, caffeine, and other drugs for 24 h, and not to eat for at least 1 h prior to testing.

The incremental exercise tests were performed in 3-min stages. Most subjects began at a power output of 40 W, which was then increased by 40 W per stage. Previous work with experienced male cyclists(19,20) led us to modify their protocol to reduce the total duration of their tests. They began at 80 W for the first stage, proceeded to 160 W for the second stage, and then had 40 W per stage increments thereafter. This allowed them to complete a similar number of stages as the other subjects. All subjects exercised until they reached exhaustion and under their own volition decided to stop or until they were no longer able to maintain the prescribed cadence. This was followed by 2-3 min of active cooldown.

Data collection and analysis. Heart rate was measured continuously on an automated ECG system (SensorMedics Max-1). Expired gases were collected continuously and analyzed to determine ventilation(˙VE), oxygen consumption (˙VO2), and carbon dioxide production (˙VCO2) using a metabolic cart (SensorMedics 2900c), whose O2 and CO2 analyzers were calibrated prior to each test against known gas concentrations and whose ventilation meter was calibrated at least once per day against a 3.0-1 syringe. Peak oxygen consumption was defined as the highest ˙VO2 obtained over any continuous 60-s time period, provided respiratory exchange ratio (RER) was ≥ 1.10. Subjects not meeting this criterion on either of their tests were excluded from the study.

˙VO2peak was predicted from data collected during the submaximal portion of the incremental tests, as has been done previously(10,15). The particular method for predicting˙VO2peak from the submaximal data was similar to that in the 5th edition of the ACSM guidelines (2). Specifically, any stage that was completed with a heart rate less than 85% of the individual's age-predicted maximal heart rate (i.e., 0.85 × (220-age)) was considered to be submaximal. For each subject and at each cadence, a linear regression was performed on the heart rate values recorded during the last 15 s of these stages with their corresponding power outputs, and the regression was extrapolated to the age-predicted maximal heart rate (220-age) to yield an estimate of maximal power. Estimated maximal power was then used to calculate predicted ˙VO2peak from the ACSM metabolic equation, i.e.,˙VO2peak in ml·min-1·kg-1 = 3.5 ml·min-1·kg-1 + 12.24·(power)·(BW-1), where power is in watts and body weight is in kilograms (1,2).

Statistics. Differences between subject characteristics (age, height, weight,% fat) were determined by independent Student'st-tests. Differences between the physiological responses of the two groups of subjects to maximal exercise (actual ˙VO2peak, HRpeak, RERpeak) under the two cadence trials were determined by 2 × 2 ANOVA with repeated measures on one factor (cadence). Linear regressions were performed on the predicted versus actual values of˙VO2peak. Pearson r correlation coefficients and standard errors of the y estimate are reported. The significance of the correlation coefficients was judged by Student's t-test. The degree to which the method for predicting ˙VO2peak over- or under-predicted the actual value of˙VO2peak was determined by calculating the% difference for each individual (i.e., (predicted vs actual)/actual) and then determining the mean% difference for each cadence. Significance for all tests was set at an alpha level of 0.05. Mean values are reported with their standard error.


As seen in Table 1, there were no significant differences in the subject characteristics between the EXP and NEXP groups.

All subjects completed at least two stages of each test submaximally, i.e., with heart rates no more than 85% of age-predicted maximum. The average number of submaximal stages completed was: for NEXP, 3.8 ± 0.2 at 50 rpm, and 3.6 ± 0.2 at 80 rpm; for EXP, 3.8 ± 0.2 at 50 rpm, and 3.9± 0.2 at 80 rpm.

There were no differences between the 50 rpm and 80 rpm trials in actual˙VO2peak, HRpeak, or RERpeak, for either the EXP or NEXP subjects (Table 2). EXP subjects had significantly higher values of actual ˙VO2peak and HRpeak than did NEXP subjects, while there was no difference in RERpeak between groups(Table 2). The peak heart rates given inTable 2 represent the following percentages of age predicted maximum heart rates (based on 220-age): for NEXP, 92.5 ± 1.0% at 50 rpm, and 92.8 ± 0.9% at 80 rpm; for EXP, 95.8 ± 1.1% at 50 rpm, and 94.7 ± 1.1% at 80 rpm.

The correlation between actual and predicted ˙VO2peak at 50 rpm for all subjects, as seen in Figure 1, was 0.79, with a SEE of 8.2 ml·min-1·kg-1. Results for 80 rpm were similar (Fig. 2), with r = 0.81, and SEE = 7.4 ml·min-1·kg-1. Both correlations were statistically significant (P < 0.001).

For EXP subjects alone, the correlations between actual and predicted˙VO2peak were r = 0.68, SEE = 8.2 ml·min-1·kg-1 at 50 rpm; r = 0.73, SEE = 7.5 ml·min-1·kg-1 at 80 rpm. For NEXP subjects alone, the correlations between actual and predicted ˙VO2peak were r = 0.65, SEE = 5.8 ml·min-1 ·kg-1 at 50 rpm; r = 0.65, SEE = 7.2 ml·min-1 ·kg-1 at 80 rpm. While there was a trend for a higher correlation coefficient at 80 vs 50 rpm for the EXP subjects, this was not statistically significant.

Predicted values of ˙VO2peak overestimated the actual values by similar amounts in EXP and NEXP subjects. For each cadence, the average overestimation for all subjects was 29.7 ± 2.8% at 50 rpm and 27.5± 2.7% at 80 rpm.


The nearly universal use of 50 rpm for fitness testing on cycle ergometers appears to have a historical basis. One of the earliest studies of bicycling economy (8) found that a cadence of 33 rpm yielded the lowest oxygen consumption over the range of 10 to 120 rpm. However, only one subject was tested. Other early studies generally found the most economical cadence to be in the range of 50 to 60 rpm (5). In the 1950's, the Astrands chose to use a cadence of 50 rpm for their submaximal fitness test (3). Since that time, most researchers have used a 50 rpm cadence for bicycle fitness tests(11,13,15,18). Fox(10) was the only exception, using a cadence of 60 rpm in 1973.

Recently, a number of studies have reexamined the question of cycling economy and cadence. Hagberg et al. (12) found that a cadence of 91 rpm was most economical among trained cyclists. Coast and Welch(6) demonstrated that cycling economy varied with power output. Among their trained subjects, 50 rpm was most economical at a power output of 100 W, while 80 rpm was the most economical at 300 W. Swain and Wilcox (19) tested the applicability of these findings by comparing the economy of uphill cycling at 84 rpm to that at a much lower cadence (41 rpm) used by cyclists and found the 84 rpm to be more economical. Given the findings of these studies of cadence, the sole use of 50 rpm for fitness testing seems questionable.

The current study has demonstrated that the prediction of˙VO2peak by submaximal bicycle testing has equal validity at 50 and at 80 rpm when employed across a wide range of subjects. Among those subjects with cycling experience, there was a trend for better validity (higher correlation coefficient) at 80 rpm; however, contrary to our hypothesis, this did not reach statistical significance. These subjects were moderately trained, not highly conditioned, athletes. Among subjects with no cycling experience, correlation coefficients were the same at 50 and at 80 rpm, although there was a lower SEE at 50 rpm. Thus, we conclude that cadences of either 50 or 80 rpm may be used for fitness testing purposes with approximately equivalent results.

A second finding was that the method used to predict ˙VO2peak significantly overestimated the actual ˙VO2peak values by approximately 28% on average. This is similar to the 26% overestimation in the recent study of Griewe et al. (11). Several earlier studies of submaximal cycle ergometry obtained better results(3,10,13,15,18). The reason for the overprediction in the current study and that of Griewe et al. appears to be the use of the ACSM's methodology of extrapolating submaximal heart rates to a predicted maximal power and then calculating a ˙VO2 from that power.

Griewe et al. (11) identified several possible reasons for the overestimation in the method they employed (ACSM's guidelines, 4th edition). These include the use of stages which were only 2 min in duration, the termination of the test at a relatively low intensity (only 70% of age-predicted maximal heart rate), possible error in the conversion factor of power to ˙VO2 (12.24 ml·min-1·W-1) which has never been validated at high workloads, and the fact that most individuals are unable to attain a heart rate of 220-age while performing cycle ergometry (11).

The ACSM modified two of these points in its 5th edition of the guidelines(2): stages were increased to 3 min in duration, and the termination criterion was raised to 85% of age-predicted maximal heart rate. These features were incorporated in the protocol of the current study, but the accuracy was virtually the same as that of Griewe et al. The ACSM's current guidelines differ from the protocol in the current study regarding the attainment of a steady-state heart rate. Specifically, the ACSM states that if the heart rates at the end of the second and third minutes are not within 6 beats·min-1, then the stage should be extended. This was not done during the collection of data for this study (which began prior to the publication of the ACSM's new guidelines). As a posthoc analysis, we reviewed all of our data and eliminated stages that did not meet this steady-state criterion. Then, we eliminated subjects who did not reach steady state on their highest submaximal workload (that which approached 85% of age-predicted maximal heart rate) and at least one other lower workload for both the 50 and 80 rpm tests. This reduced the number of subjects from 58 to 40 but did not improve the correlations or reduce the overestimation of˙VO2peak. From this analysis, we find that the minor changes in the ACSM protocol from the 4th to the 5th edition do not improve the accuracy of the test over that reported by Griewe et al. (11).

Griewe et al. (11) pointed out that peak heart rate during cycling is lower than 220-age and is one source of error. In the current study the peak heart rate of the EXP subjects was significantly higher than that of the NEXP subjects, and thus closer to 220-age. This would suggest that the prediction of ˙VO2peak in the EXP subjects would be closer to actual ˙VO2peak than in the NEXP subjects. However, the overestimation of ˙VO2peak was equal in the two groups. Another factor to consider is the assumption of linearity for the heart rate/power relationship. Recently, several investigators demonstrated that a majority of endurance trained subjects exhibit a heart rate deflection at moderate to high intensities, a point where the slope of the heart rate/power relationship is reduced (7,9,14,17,20). Since the extrapolation method used by the ACSM assumes a constant slope that is based primarily on data obtained at low to moderate intensities, this would overestimate maximal power in individuals with a heart rate deflection.


Under the conditions of the current study, cadences of 50 and 80 rpm are equally valid for the purposes of predicting ˙VO2peak from submaximal cycle ergometry. It is suggested that professionals performing fitness tests use either cadence, depending on the background or preference of the client. The method of predicting ˙VO2peak used by the ACSM overestimates actual ˙VO2peak, and a return to methodology involving fewer assumptions should be considered. More than 40 years ago, the use of a single 6-min stage produced fairly accurate results(3). Future research may determine what workload for the 6-min stage and what protocol for achieving it is most appropriate for use with subjects of any experience level.

Figure 1-Linear regression at 50 rpm for all subjects. Regression equation is: predicted ˙VO2peak = 12.9 + 0.946 × actual˙VO2peak.
Figure 1-Linear regression at 50 rpm for all subjects. Regression equation is: predicted ˙VO2peak = 12.9 + 0.946 × actual˙VO2peak.
Figure 2-Linear regression at 80 rpm for all subjects. Regression equation is: predicted ˙VO2peak = 10.6 + 0.991 × actual˙VO2peak.
Figure 2-Linear regression at 80 rpm for all subjects. Regression equation is: predicted ˙VO2peak = 10.6 + 0.991 × actual˙VO2peak.


1. American College of Sports Medicine. Guidelines for Exercise Testing and Prescription, 4th Ed. Philadelphia: Lea & Febiger, 1991, pp. 5-7, 42, 296.
2. American College of Sports Medicine. ACSM's Guidelines for Exercise Testing and Prescription, 5th Ed. Baltimore: Williams & Wilkins, 1995, pp. 65, 70, 278-279.
3. Astrand, P. O. and I. R. Rhyming. A nomogram for calculation of aerobic capacity (physical fitness) from pulse rate during submaximal work. J. Appl. Physiol. 7:218-221, 1954.
4. Banister, E. W. and R. C. Jackson. The effect of speed and load changes on oxygen intake for equivalent power outputs during bicycle ergometry. Int. Z. Angew. Physiol. Einschl. Arbeitsphysiol. 24: 284-290, 1967.
5. Coast, J. R. Optimal pedaling cadence. In:High-Tech Cycling, E. R. Burke (Ed.). Champaign, IL: Human Kinetics, 1996, pp. 101-116.
6. Coast, J. R. and H. G. Welch. Linear increase in optimal pedalling rate with increased power output in cycle ergometry. Eur. J. Appl. Physiol. 53:339-342, 1985.
7. Conconi, F., M. Ferrari, P. G. Ziglio, P. Droghetti, and L. Codeca. Determination of the anaerobic threshold by a non invasive field test in runners. J. Appl. Physiol. 52:869-873, 1982.
8. Dickinson, S. The efficiency of bicycle pedalling as affected by speed and load. J. Physiol. (Lond.) 67:242-255, 1929.
9. Droghetti, P., C. Borsetto, I. Casoni, et al. Non invasive determination of the anaerobic threshold in canoeing, cross-country skiing, cycling, roller and ice skating, rowing, and walking. Eur. J. Appl. Physiol. 53:299-303, 1985.
10. Fox, E. L. A simple, accurate technique for predicting maximal aerobic power. J. Appl. Physiol. 35:914-916, 1973.
11. Greiwe, J. S., L. A. Kaminsky, M. H. Whaley, and G. B. Dwyer. Evaluation of the ACSM submaximal ergometer test for estimating˙VO2max. Med. Sci. Sports Exerc. 27:1315-1320, 1995.
12. Hagberg, J. M., J. P. Mullin, M. D. Giese, and E. Spitznagel. Effect of pedaling rate on submaximal exercise responses of competitive cyclists. J. Appl. Physiol. 51:447-451, 1981.
13. Hartung, G. H., R. J. Blancq, D. A. Lally, and L. P. Krock. Estimation of aerobic-capacity from submaximal cycle ergometry in women. Med. Sci. Sports Exerc. 27:452-457, 1995.
14. Kuipers, H., H. Keizer, T. de Vries, P. van Rijthoven, and M. Wijts. Comparison of heart rate as a non-invasive determinant of anaerobic threshold with the lactate threshold when cycling. Eur. J. Appl. Physiol. 58:303-306, 1988.
15. Mastropaolo, J. A. Prediction of maximal O2 consumption in middle-aged men by multiple regression. Med. Sci. Sports Exerc. 2:124-127, 1970.
16. Pollock, M. L., D. H. Schmidt, and A. S. Jackson. Measurement of cardiorespiratory fitness and body composition in the clinical setting. Comp. Ther. 6:12-27, 1980.
17. Ribeiro, J., R. Fielding, V. Hughes, A. Black, M. Bochese, and H. Knuttgen. Heart rate break point may coincide with the anaerobic and not the aerobic threshold. Int. J. Sports Med. 6:220-224, 1985.
18. Siconolfi, S. F., E. M. Cullinane, R. A. Carleton, and P. D. Thompson. Assessing ˙VO2max in epidemiologic studies: modification of the Astrand-Rhyming test. Med. Sci. Sports Exerc. 14:335-338, 1982.
19. Swain, D. P. and J. P. Wilcox. Effect of cadence on the economy of uphill cycling. Med Sci. Sports Exerc. 24:1123-1127, 1992.
20. Swain, D. P., J. P. Wilcox, and T. A. Shepherd. Does heart rate deflection represent the “anaerobic threshold”?Med. Sci. Sports Exerc. 24(Suppl.):123, 1992.


©1997The American College of Sports Medicine